Glossary
CMDB
A CMDB (Configuration Management Database) is a centralised database that lists an organization's IT assets and their relationships. It is at the heart of IT mapping and ITSM governance.
Updated
Definition of CMDB
CMDB stands for Configuration Management Database. It is a centralised repository describing all the configuration items (CIs) of an information system and the relationships between them.
The concept originates from the ITIL framework (Information Technology Infrastructure Library), first published in the 1980s by the UK's CCTA agency, later taken over by AXELOS. ITIL defines the CMDB as the central component of the Service Asset and Configuration Management process, one of the processes of the Service Transition phase.
What does a CMDB contain?
A CMDB lists configuration items, which can be physical or logical assets. Physical assets include servers, workstations, network equipment and IoT sensors. Logical assets include applications, business services, databases, virtual machines, containers and Kubernetes namespaces.
Each CI carries attributes: unique ID, name, type, owner, environment (production, staging, development), criticality, status (live, maintenance, retired), version, acquisition date, attached support contract.
The added value of a CMDB lies not just in the list of assets, but in the modeling of relationships between them: which app runs on which server, which business service depends on which app, which data flow circulates between which components. These relationships are what makes impact analysis possible during an incident or change.
What is the difference between a CMDB and an inventory?
Confusion between CMDB and inventory is common. The distinction is about modeling depth and purpose.
An inventory is a flat list of assets: how many servers you have, in which datacenters, on which OS version. Useful for budget management, support contract tracking or license compliance. But an inventory says nothing about how assets interact.
A CMDB adds the relationships: this server hosts these three apps, consumed by this HR service, itself critical to the payroll process. This modeling makes the CMDB usable for impact analysis, change management, business continuity planning and regulatory compliance (notably NIS2 and ISO 27001).
Why legacy CMDBs often fail
The idea of an exhaustive, up-to-date CMDB has been appealing for thirty years, but the failure rate of CMDB projects remains high. Causes are structural.
First cause: manual maintenance. A CMDB fed by hand by operations teams quickly drifts towards obsolescence. Daily infrastructure changes (cloud provisioning, CI/CD deployments, updates) saturate any human entry capacity.
Second cause: scope too broad. Trying to inventory every asset down to network cables drowns useful information and demobilises teams. Modern CMDBs adopt a federation approach: dynamically pull CIs from source-of-truth tools (application CMDB from monitoring, infrastructure CMDB from IaC, application CMDB from source code).
Third cause: no clear business use. A CMDB that only serves itself is never maintained. What justifies keeping a CMDB up to date is concrete use: incident impact analysis, change simulation, compliance audit.
An incomplete CMDB: what to do, and what IT mapping adds
An incomplete CMDB is still a useful source. It often holds what no technical export knows: who owns an application, how critical it is, the support contract, the business service it delivers. What it lacks are the machines created without a ticket, the versions actually installed and the flows actually open.
The approach that holds is to compare it with your other sources instead of rebuilding it: the hypervisor export says what runs, the cloud accounts say what was created, the firewall rules say what talks to what. Each gap becomes a precise question for a named person, not a data-entry project.
That is the job of IT system mapping drawn from sources. Uncia reads your CMDB as it is, read-only, reconciles it with your other sources and shows where they diverge. It does not replace the CMDB and writes nothing to it: corrections happen in your own tool.
Frequently asked questions
Is ITIL mandatory to have a CMDB?
No, ITIL is a best-practice framework, not a standard. You can operate a CMDB without formalizing the full ITIL framework. In practice, many organizations adopt the concepts (CI, relationships, change process) without the full administrative overhead.
What is the difference between a CMDB and an architecture repository?
A CMDB is centered on configuration items and their operational use. An architecture repository (often called EA, Enterprise Architecture) includes the business vision, functional capabilities and evolution roadmap. The two overlap on applications and their dependencies, and they are usually kept in separate tools.
How much does it cost to set up a CMDB?
It depends mostly on how much is entered by hand. A hand-fed CMDB needs constant upkeep, which stops as soon as the team is busy elsewhere. Starting from existing sources (hypervisor, cloud, firewall) reduces that share: the base is read again at every collection, and manual entry is limited to what no source knows.